Minutes for 6TISCH at interim-2016-6tisch-13
IPv6 over the TSCH mode of IEEE 802.15.4e
||Minutes for 6TISCH at interim-2016-6tisch-13
Minutes, 23 September 2016 interim, 6TiSCH WG
Note: timestamps in PDT.
Date: 7-8am Pacific:
Recording password: XsN3nVq8
Taking notes (using Etherpad)
Pascal to fix the milestone ( draft name)
Thomas to ask consensus on rcv initiated 6P transaction and cell suggestion
Xavi to start a thread for consensus pn number of cells in response
Xavi to propose to carry the link option in teh 6top protocol
Approval minutes last call
working with 802 [5min]
6top protocol [10min]
agreeing on action items, milestones and next steps
Receiver Initiated 6P Transactions
Receiver 6P Cell Suggestion
dynamic number of cells in reply
Next steps for SF0 [30min]
status and next steps
Cell allocation Scheme
[07.05] (expected 07:05) Administrivia [3min]
Meeting starts, agenda Bashing
agenda and minutes of last call approve
udpate of charter . renaming 6top protocol from sublayer.
date of december to submit 6top and sf0 to the iesg. Would be desirable
to send the documents by the next ietf meeting Thomas: adding to the
charter 4th etsi plugtest in the next year july ietf meeting (2017)
[implies no plugfest in Feb.2017? I guess no] paging dispatch document
submitted to IESG. reviewed by all the areas. Good feedback. routing
dispatch is the real 6lorh, this is reviewed and will be pushed to the
IESG routing dispatch and paging dispatch drafts should update
rfc6550-51-52-53. minimal draft is in the internet area review. Hanging
there. Michael: every second tuesday meeting of the security design
[07.??] (expected 07:08) working with 802 [5min]
An IEEE and IETF meeting in Paris happened last week. discussion about
things that are crossing both groups. 6TiSCH IEs in 6top. The IETF is
requesting to the IEEE. 6lo is requesting an Ethertype to the IEEE.
There will be an official ethertype like IPv6.
[07.??] (expected 07:13) 6top protocol [10min]
Thomas: draft is almost ready.
2 ideas that are floating. Receiver initiated transactions. Receiver
cell ... allow the node to limit the numbr of cells if not enough room
in the packet due to IE Xavi: I'm handling that. 1st idea (rcv
initiated transaction): consensus? If yes I can handle it. Thomas:
action item to get consensus on that on the ML Xavi: no strong opinion.
Mechanism looks simple; again need consensus. Pascal: 2 action items
for Thomas, one for Xavi Xavi: also need to validate the draft on
missing ack. check that there is a timeout on the 2 way transaction?
Thomas: only missing point is security, making progress. Very close to
a full package
[07.??] (expected 07:23) Next steps for SF0 [30min]
Thomas: reviewed SF0. Needs work; This is a key piece, the default
brains. Thomas: slide 18 identifies questions, Diego is the only author
present Need level of evaluatuion / proof to get std track. Pascal:
need to track the issues as tickets on the IETF tracker, please use
different threads for each ticket Pascal to handle the issue tracking
in the IETF tracker to track the threads to resolution. many acronyms
in the draft makes it confusing. Simplify, the acronyms for the data
flows. outband bandwidth is not calculated correctly. Check the
formulation. whitelist/blacklist is like a not, to say if we want to
reserve cells out of the available (blacklist). Whitelist indicates
just the list of cells that wants to be allocated. In the metadata this
is specified. timeout calculation: is calculated as the number of cells
until the next opportunity to transmit. This timeout should be larger
than the normal delay instead as we need to take into account the PDR.
tengfei: the timeout should be related to the backoff period. Not to
the PDR. 6top protocol packets are sent on shared slots only? Pascal it
should be more flexible, 2 extra slots per peer is sometimes too much,
somtimes not enough. SHould be able to use chared slots. Thomas: it is
up to the minimal configuration to decide the amount of shared cells.
Pascal: That works for a minimal only solution, because we can tune the
size of the slotframe. But here it will be dictated by eg how many
slots are needed for all these peers, and one shared may not be enough.
Suggestion is that SF0 also allowates shared slots taht teh parent
exposes to selected children. Thomas: SF0 is a simple mechanism that is
not aware of parent / child, keep it simple. Also the concept of shared
slots is that all my neighbors see it. Pascal: a shared slot is used by
everybody that knows about it. It is not a transitive proerty anyway.
Xavi: proposal to use the link option when a cell is requested in 6top.
Action to add linkoption to the 6p draft. Pascal: agrees to have the
linkoption but by now we do not use it in the sf0
[08.05] (expected 07:53) AOB [2min]
[08.08] (expected 07:55) Meeting ends